2017 Annual Meeting
(716e) CFD-DEM Modeling on Exascale Computer Architectures
Authors
An opportunity to address that challenge is becoming available through the Exascale Computing Project (ECP), which began in September 2016. ECP is a collaborative effort of two US Department of Energy organizations â the Office of Science and the National Nuclear Security Administration. ECP was established to accelerate delivery of a capable exascale computing system that integrates hardware and software capability to deliver approximately 50 times more performance than the most powerful supercomputers in use today in the US. ECPâs work encompasses applications, system software, hardware technologies and architectures.
One of the applications being developed by ECP is MFIX-Exa, which has the goal of conducting CFD-DEM simulations on exascale architectures. The 10-year target of MFIX-Exa is a code that will simulate several minutes of the operation of a multiphase reactor, containing around a hundred billion particles, in less than twenty-four hours of wall time. MFIX-Exa will be developed from the widely used, open-source code MFIX-DEM and a new software framework called AMReX, also being developed by ECP. AMReX will use block-structured adaptive mesh refinement algorithms to solve systems of partial differential equations with complex boundary conditions on exascale architectures. The capability of MFIX-Exa will be demonstrated by modeling a ~1 MWe chemical looping reactor. MFIX-Exa will also run on conventional architectures, and a preliminary release is expected in the next three years.
The capability to model billions of reacting particles will make it possible to solve an entirely new class of problems. The simulations of reactors at the scale of ~1 MWe will surpass experiments in cost, speed, and knowledge gained. CFD-DEM simulations will also provide currently-unavailable data for validating coarse-grained models (such as two-fluid or particle-in-cell) required for simulating larger, industrial-scale reactors.